I just fired up my Windows 7 box and tried the PropBOE with the SimpleIDE that is installed on that box, I get the same results. So, I guess I will commit and say that I have a bad PropBOE. Now, I am wondering, did I do something unusual or did I really get a bad board. If it was my fault, I sure as heck do not want to repeat that with a new board. Since these boards were intended for use in a computer lab setting, I would make the assumption that you would not have to practice exceptional handling procedures.
If it is a bad board I wonder why LMM works and CMM doesn't. Maybe it's a bad bit in hub RAM, or it could be something else. Is there any diagnostic code available to help determine if a chip is good or bad?
Here is a program that you can use to test starting up to 7 cogs. I would like to see if you have any LMM (or CMM) failures with this code on your board. This is basically an expansion of your program renamed to ncogs.c
Change the LED pin with the "pin" variable. Change the number of cogs to start with the "count" variable. Both are at the top of the file.
I just ran your little program, first I ran it in LMM just to see what it is supposed to do, and it ran as expected. As I switched over to CMM mode, it just fell apart, the LED stayed lit for awhile, and then the screen went blank, so it did not work as expected. So I will call it quits with this board, something is definitely wrong.
I should be getting some diagnostic code soon from Parallax that we can use for testing your board.
I'll probably have to make some custom binaries. Doubt anything will be available tonight.
Thinking along the lines of diagnostic tools, has anybody come up with a program to check the processors RAM? I would image such a program would have to be written in PASM, or maybe C would be able to do something like that? Now, if there is/was such glitch in RAM, how would that affect the programs that are written in Spin? As the production rate of the Propeller increases so do the odds of having glitches in the processor itself.
In another thread, in another forum, a forum member made a very subtle comment about "the troubles" of PropGCC, well it looks like "the troubles" might be with the processor and nothing to do with PropGCC, in fact, because of PropGCC, the possibility of a processor problem might have been revealed. I did not start this thread to malign PropGCC, or to provide an atmosphere for anybody else to take cheap shots, I was just posting the occurrences with my board. In my limited experience with the C language, I have found that working with SimpleIDE and PropGCC, to be very gratifying, enough said, now I want to get back to programming.
Thinking along the lines of diagnostic tools, has anybody come up with a program to check the processors RAM?
Attached is the Parallax HUB RAM test .binary - load it with Propeller-Tool. Set your serial connection to 115200 bps.
It will print this in a few seconds if all goes well:
I just ran the Hub_RAM_Tester_1.0.binary, and got the results below. Dave Hein is one sharp programmer, he was onto something when he mentioned HUB RAM problem(s). So, I wonder if the processor has other problems. Again, SimpleIDE and PropGCC is one excellent software package, hats off to jazzed and the team.
Memory and processor problems were pretty common back in the 70s and 80s. That's why most S100 memory boards had 9-bit memory -- 1 bit was for parity. The problem you were seeing reminded of those problem. Usually strange issues like that are due to uninitialized variables, stack overflows or accessing beyond the bounds of an array. It's been a while since I've seen a hardware problem.
@Daniel Harris, I got my replacement board today, this board is returning to Ken Gracey, by his request.
If their are no more tests that anybody can think of, I will put the board with a glitch, in the mail, tomorrow morning. So, all the curious people will still have a chance to come up with something, for me to try, for the rest of today, remember I am EST, and I tend to quit my hobby activities very early evening.
Unfortunately, I do not have a Spin program that uses FullDuplexDriver, in a way to see if there would be any erroneous data being transferred. I figure a Spin program that would be written in such a way where it would use Cog1, could reveal if Spin programs still function correctly with an existing bad Cog.
Just a quick follow-up on Ray's board. P9 was indeed bad, so this board is now happily sitting in Tech Support where it is destined for future uses [that don't require P9].
Comments
Ray
Could you try a test that adds cogs in LMM mode?
I'll post a test for you in about an hour.
Ray
Here is a program that you can use to test starting up to 7 cogs. I would like to see if you have any LMM (or CMM) failures with this code on your board. This is basically an expansion of your program renamed to ncogs.c
Change the LED pin with the "pin" variable. Change the number of cogs to start with the "count" variable. Both are at the top of the file.
Thanks,
--Steve
Ray
I'll probably have to make some custom binaries. Doubt anything will be available tonight.
In another thread, in another forum, a forum member made a very subtle comment about "the troubles" of PropGCC, well it looks like "the troubles" might be with the processor and nothing to do with PropGCC, in fact, because of PropGCC, the possibility of a processor problem might have been revealed. I did not start this thread to malign PropGCC, or to provide an atmosphere for anybody else to take cheap shots, I was just posting the occurrences with my board. In my limited experience with the C language, I have found that working with SimpleIDE and PropGCC, to be very gratifying, enough said, now I want to get back to programming.
Ray
Attached is the Parallax HUB RAM test .binary - load it with Propeller-Tool. Set your serial connection to 115200 bps.
It will print this in a few seconds if all goes well:
Cog 0: PASS
Cog 1: PASS
Cog 2: PASS
Cog 3: PASS
Cog 4: PASS
Cog 5: PASS
Cog 6: PASS
Cog 7: PASS
If there is a glitch in HUB RAM, it is possible that some things wont work and others will.
There are a few other tests, but this one should get you started.
Ray
Would you PM me your contact info (email or phone). I'd like to get that board back for additional testing. Of course we'll replace it
Thank you!
Daniel
If their are no more tests that anybody can think of, I will put the board with a glitch, in the mail, tomorrow morning. So, all the curious people will still have a chance to come up with something, for me to try, for the rest of today, remember I am EST, and I tend to quit my hobby activities very early evening.
Unfortunately, I do not have a Spin program that uses FullDuplexDriver, in a way to see if there would be any erroneous data being transferred. I figure a Spin program that would be written in such a way where it would use Cog1, could reveal if Spin programs still function correctly with an existing bad Cog.
Ray
I'm glad the HUB_RAM_Tester program caught the issue.
Once Parallax has the bad board, they will do more testing.
This is the first time I've ever heard of a Propeller chip with a memory error.
Good luck with the new board.
-Phil